Abstract Monitoring changes in groundwater storage (ΔGWS) is critical to assess sustainability of groundwater use under climate variability. Satellite gravimetry from the GRACE missions is used to infer ΔGWS by deducting changes in monitored or modeled water components from GRACE‐derived changes in total water storage (ΔTWS). As a residual parameter, ΔGWS is highly sensitive to arithmetic inconsistencies and uncertainties in both GRACE data and model‐derived inputs. Here we present a framework to evaluate the physical plausibility of GRACE‐derived ΔGWS estimates across 37 large global aquifer systems. The results show that the proportion of plausible ΔGWS estimates per realization, derived from multiple GRACE products and land surface model combinations, varies from 60%. Exclusion of implausible estimates improved correlations between ΔGWS and ΔTWS substantially (r ≥ 0.9, p‐value <0.05) in most aquifers (31/37) and with in situ observations in the Bengal Basin (r = 0.8, p‐value <0.05).

Read original article